This relates generally to compressing data before storing in a memory.
Memory compression saves power and bandwidth in processor-based systems including central processing units, graphics processing units and other applications. To allow for better compression ratios, different compression techniques may be used based on the typical usage of buffers, for example in connection with different types of content such as media and graphics.
Furthermore since compression may be lossless, there is a need for metadata that indicates whether or not the data actually underwent compression. This metadata may be available for successive cachelines. The number of cachelines grouped together depends on the compression algorithm. Generally when a memory access occurs, it is directed to the actual data itself and thus there is the problem of how to access the compression metadata.